 Philosophy, video games, and the single most important question in neuroscience. How's that for a brain teaser? Hey, thunk fans. Sorry this video is a little bit late. I sometimes get crippling migraines, and I had one at exactly the wrong time yesterday. It was thematically appropriate, as you'll see, but at the time it wasn't very amusing. Anyways, thank you for your patience. Let's get started with physicalism. Physicalism is the idea that everything which exists is physical in nature. There are a few different ways of interpreting what that means, but in short, physics or bust. Physicalists, sometimes called materialists, believe that for everything that's real, even things that have very weird or complex properties, those properties must supervene or be entirely entailed by physical processes. That in the end, they are due solely to the interactions of matter and energy. For example, your computer is doing some incredibly complicated stuff right now. Memory management, communication with other computers, displaying information. The complex behavior that we see computers doing, like playing a YouTube video, supervenes on the behavior of that computer's hardware, like hard drive spinning or transistor switching, which in turn supervenes on the material properties of the atoms that make up that hardware, like copper and silicon. If you asked why is the computer doing that, there are different levels of explanations that you might give, like, well, because I clicked on that link, or, well, because YouTube servers are streaming packets to this computer to display over the internet. But for physicalists, all those answers can ultimately be boiled down to, well, because physics. Physicalism is discussed frequently in the philosophy of mind, mainly because it lies contrary to a popular belief that human minds have some existence independent of the material world, that there's something special going on outside of the raw chemistry of neurons and cells. Physicalists don't deny that minds are incredibly complex and amazing. They can still believe in things like love and intelligence and all that, but they do believe that all those phenomena ultimately supervene on standard interactions of matter. For any mental state, the argument goes, whether you're feeling a little bit hungry or thinking about what you're going to have for lunch, that mental state is the direct consequence of the velocities and positions of particles in your body. If you were to build an exact physical copy of your brain and feed it the exact same input, then it would feel and behave exactly the way that yours does right now. It's essentially the default position for all scientific investigation – cognitive science, psychology, neuroscience. All of these fields more or less implicitly assume that your mental state is identical with the physical structure and behavior of matter in your brain. Well, because physics. As anyone who's benefited from that science can tell you, whether they're getting meds for some sort of psychiatric disorder or recovering from a stroke, that assumption is definitely useful, both for discovering new truths about how brains and minds work and for rendering aid when they're suffering. And if you take it a step further, that's a powerful motivation to continue trying to understand the physical rules by which your brain operates. If physicalism is true, if the operation of your hardware is the beginning and end of whatever your mind does, it's the beginning and end of your entire world. Of course, we're a long, long way from building a perfect copy of your brain to prove or disprove cognitive physicalism for sure. A hundred billion neurons, several hundred trillion connections in a constantly shifting, self-modifying network. That makes it one of the most complicated, intricate, and sensitive mechanisms that we've ever discovered. I have trouble plugging in USB cables the right way. Despite this complexity, neuroscience, the study of the physical structure and function of the nervous system, has uncovered some really amazing insights. Like, we can use deep brain stimulation to simply switch off tremors in some people. We've discovered a specific brain region that's responsible for feelings of religious awe and rapture. But the holy grail of neuroscience is decoding exactly how human brains process information. When I show you a picture of a cute puppy, what exactly happens in here? What precise sequence does this vast network of neurons follow between receiving this pattern of colors and shapes from your eye and making you say, oh, I mean, we can make some educated guesses comparison of the image with memory, evaluation of its semantic content, emotional response. But how would you program that sequence using a hundred billion neurons? In other words, how do you make a functional brain? It's, shall we say, a tough question. Adding to the magnitude of this problem is just how hard brains are to research in the first place. They have this annoyingly large laundry list of operational requirements that aren't really conducive to research. I mean, look at this. Must have blood flow. Must not be exposed to excessive pain. Must be inside the skull and attached to a living body. It just goes on and on like this. Of course, neuroscientists are very clever people, and they've worked hard to develop analytical techniques on the data that they can get so they can try and work out what the hell's going on in there. For example, examining behavior before and after a lesion in a particular area of the brain might give us some idea of what that area is responsible for. Mapping out the links between neurons might give us some sort of wiring diagram which might tell us how it's structured. Many of these techniques aren't usable on humans, mainly because causing permanent brain damage or killing people is generally frowned upon in most research fields. That lack of information on the one thing that we're trying to figure out has been very frustrating for many researchers. We can get a very detailed picture of what's going on in animal brains or a very vague idea of what's going on in human brains, but neither one is very helpful. Crucially, we don't really know if any of these techniques would work. Even if we did find a way to use them safely on people, if we found some sort of high-resolution, unobtrusive way to record neural activity in someone's head, would we really be any closer to decoding the brain? In their paper, could a neuroscientist understand a microprocessor? Some researchers put these neuroscience techniques up against some hardware that we can get a ton of information about. Microchips. Specifically, the Moss 6502 microprocessor, which was used in the Commodore 64 and the Atari. This is a well-understood physical system, with some complexity in how its components are put together and a set of very predictable behavioral responses to certain inputs. For the purposes of the study, they chose some behaviors that you're probably familiar with. Donkey Kong, Space Invaders, and Pitfall. No, really. The Moss 6502 is simpler than brains are. It has about 28 million times fewer transistors than brains have neurons. But those neuroscience techniques for decoding how brains process information should work wonders on such a simple problem, right? Unfortunately, not so much. Although the 11 techniques tried sometimes uncovered interesting patterns in the behavior of the chip, they never really provided any help figuring out the rules of its operation. No matter how you analyze the unlimited and 100% complete and perfectly recorded data generated by doing various things to it, there wasn't really anything in that data that said, this is what's going on here. Now, obviously, these results are to be taken with a grain of salt. Brains and processors might be analogous. They might even function on similar principles. But trying to pretend that techniques used on one could safely extend to the other without any issues is kind of a stretch. It's a little like saying, look, my RC car looks just like your Chevy, so I know exactly how to fix it. Now, have you tried changing the batteries in the controller? However, the results do raise some worthwhile concerns about the ultimate efficacy of these techniques for understanding the human brain, even if we did have unlimited access to a slew of 100% functional and transparent brains to use them on. These concerns have resonated with researchers who have been questioning the underlying assumptions necessary for these methods. For example, take connectome modeling, the wiring diagram approach. Say that we did manage to map out every single one of the multiple trillion connections between every single pair of neurons. What then? I mean, Google's image recognition algorithm is a neural network. A piece of learning software designed to do more or less the same thing that brains do, strengthen connections between some neurons if those connections work, and weaken connections between other neurons if they don't work. Google's engineers can print out the connectome of their image recognition network with incredible precision. They know exactly which nodes are connected where and how strongly. And yet, they still don't really have any idea what rules it's following to be able to tell a strawberry from a puppy. As this study illustrates, it seems likely that there's some crucial insight that's necessary for interpreting the data gathered by these techniques. I mean, it's still possible that one of them actually contains the key that we need to figure out what's going on in here. But if this paper is any indication, we're still missing an important piece of the puzzle between the information that we might gather using these techniques and the understanding that we would need to build a working brain. But believe it or not, this might be great news. Rather than sitting around frustrated that they can't gather more information about real brains, this gives neuroscientists and other researchers a way to test new methods and find out what's helpful. Maybe with this new paradigm of reverse engineering microprocessors, when we finally do get a glimpse inside someone's head, we'll be able to tell immediately what they're thinking about, especially if they're thinking about playing Donkey Kong. What do you think of physicalism? When do you think I'll be able to slip into a nice new shiny robot body? Please leave a comment below and let me know what you think. Thank you very much for watching. Don't forget to blah, blah, subscribe, blah, share. And don't stop thunking.